A self-adaptive classifier system

Hurst, Jacob; Bull, Larry

Abstract

© Springer-Verlag Berlin Heidelberg 2001. The use and benefits of self-adaptive parameters, particularly mutation, are well-known within evolutionary computing. In this paper we examine the use of parameter self-adaptation in Michigan-style Classifier Systems with the aim of improving their performance and ease of use. We implement a fully self-adaptive ZCS classifier and examine its performance in a multi-step environment. It is shown that the mutation rate, learning rate, discount factor and tax rate can be developed along with an appropriate solution/rule-base, resulting in improved performance over results using fixed rate parameters. We go on to show that the benefits of self-adaptation are particularly marked in non-stationary environments.

• Presentation Conference Type: Conference Paper (published)
• Conference Name: Advances in Learning Classifier Systems
• Start Date: Sep 15, 2000
• End Date: Sep 16, 2000
• Publication Date: Jan 1, 2001
• Publisher: Springer Verlag
• Peer Reviewed: Not Peer Reviewed
• Pages: 70-79
• Series Title: Lecture Notes in Computer Science
• Series Number: 1996
• Series ISSN: 0302-9743
• Book Title: Advances in Learning Classifier Systems
• ISBN: 9783540424376
• DOI: https://doi.org/10.1007/3-540-44640-0_6
• Keywords: artificial intelligence, mathematical logic and formal languages, computation by abstract devices
• Public URL: https://uwe-repository.worktribe.com/output/1091270
• Publisher URL: http://dx.doi.org/10.1007/3-540-44640-0_6